llms.txt

# FormulAI
> AI-Driven Formulation Intelligence for Specialty Chemicals

FormulAI is a professional-grade assistant for chemical R&D teams. It streamlines the transition from a product brief to a structured formulation concept card, balancing performance, cost, and sustainability.

## Core Capabilities
- **Generative Formulation:** Streams structured JSON from Groq API to define ingredient ranges and architectures.
- **Multi-Objective Optimization:** Evaluates trade-offs between sustainability posture, cost, and regulatory compliance.
- **Expert Review Workspace:** Provides a live environment for chemists to tune "Software-to-Steel" pipelines.
- **Regulatory Awareness:** Filters concepts based on regional regulatory flags (EU, NA, etc.).

## Technical Stack
- **Frontend:** React + Vite, styled with Tailwind CSS for a premium "cinematic" UI.
- **Backend:** Express API (Node.js) optimized for Vercel deployment.
- **Inference:** Groq API utilizing JSON mode for high-speed, structured data extraction.
- **State:** Persistence via `localStorage` for session continuity.

## Key Logic & Constraints
- **Safe-by-Design:** Prioritizes circular innovation and carbon reduction in ingredient selection.
- **Structured Output:** All formulation data must conform to a strict schema for ingredient ranges and trade-off scoring.
- **Expert-in-the-Loop:** Explicitly designed as a screening tool that requires final bench validation and SDS review by a qualified chemist.

## Project Structure
- `/src`: React components and UI logic.
- `/api`: Vercel serverless functions (entrypoint: `api/formulate.js`).
- `/server`: Core Express logic for local development.
- `.env.example`: Template for GROQ_API_KEY and model selection.